Show HN: AI decision model plays Mortal Kombat II (github.com)

A new AI decision model named Clef-flash has showcased its capabilities by playing Mortal Kombat II on the Sega Genesis, marking a significant step in how AI can harness game memory rather than pixel-based graphics for decision-making. This model was evaluated through a structured benchmark, FightBench, which allowed for frozen test scenarios where Clef-flash faced off against a Very Hard CPU opponent 15 times, emphasizing the AI's ability to respond to in-game situations rapidly and effectively. Clef-flash's performance included handling complex moves and situational queries, making it a notable advancement in AI-enabled gaming strategies. The significance of this development lies in its potential implications for AI/ML research in game theory and real-time decision-making under pressure. By analyzing game memory every 0.12 seconds and evaluating multiple potential moves based on probabilities, the model illustrates how AI can engage in strategic gameplay similar to human players. With its benchmarks providing insights into mean damage dealt, survival rates, and move efficacy, Clef-flash not only enhances our understanding of AI capabilities but also sets a foundation for future research into real-time artificial intelligence application in dynamic environments.